Premium
Model‐based methods for generating equiluminant texture targets in imaging systems
Author(s) -
Papathomas Thomas V.
Publication year - 1996
Publication title -
international journal of imaging systems and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 47
eISSN - 1098-1098
pISSN - 0899-9457
DOI - 10.1002/(sici)1098-1098(199622)7:2<78::aid-ima2>3.0.co;2-p
Subject(s) - computer science , luminance , computer vision , artificial intelligence , stereoscopy , motion (physics) , contrast (vision) , texture (cosmology) , point (geometry) , mathematics , image (mathematics) , geometry
Recently developed techniques for obtaining texture targets whose space‐average luminance is equal to a standard uniform luminance are reviewed. Such targets can be used in psychophysical experiments to obtain parameters for improving image compression algorithms based on the characteristics of the human visual system. The techniques reviewed in this article have been developed for studying apparent motion, but can be extended to cover applications in textural grouping and stereoscopic imaging systems. The point of departure of the techniques reviewed here is that properties of first‐order motion (elicited by targets that are discriminated from the background by virtue of their mean luminance difference) are distinct from those of second‐order motion (elicited by contrast‐defined targets that, on average, are equiluminant to the background). A model that accounts for most of the psychophysical experiments is one consisting of two parallel pathways for processing first‐ and second‐order motion. We present the rationale of each technique under this model, examine the expected and actual performances, and conclude with the advantages and disadvantages of each method. © 1996 John Wiley & Sons, Inc.